Creating Complex Polyacrylamide Hydrogel Structures Using 3D Printing with Applications to Mechanobiology

Abstract

Due to its favorable physical and chemical properties, including chemical inertness, low fouling by biological molecules, high porosity and permeability, optical transparency, and adjustable elasticity, polyacrylamide has found a wide range of biomedical and non-biomedical applications. To further increase its versatility, this communication describes a simple method, using readily available reagents and equipment, for 3D printing polyacrylamide hydrogels at a resolution of 100–150 microns to create complex structures. As a demonstration of the application, the method is used for creating a lab-on-a-chip cell culture surface with micropatterned stiffness, which then led to the discovery of stiffness-guided collective cell segregation distinct from durotaxis. The present technology is expected to unleash new applications such as the construction of biocompatible elastic medical devices and artificial organs.

Graphical Abstract

A simple method is introduced for 3D printing polyacrylamide to generate complex hydrogel structures. The method is then used for creating a lab-on-a-chip cell culture surface with micropatterned stiffness for testing stiffness-guided collective cell migration. This technology is expected to unleash broad applications such as the construction of biocompatible elastic medical devices and artificial organs.

Biomedical applications of polyacrylamide hydrogels include protein and nucleic acid separation,^[1] reconstructive and plastic surgery,^[2,3], contact lenses,^[3] and drug delivery,^[3,4] facilitated by such properties as chemical inertness, low fouling, high water retention, optical transparency, adjustable elasticity, and high permeability. In addition, polyacrylamide hydrogels have been used extensively in mechanobiology^[5] and in related biomedical engineering applications that involve mechanotransduction.^[6] However, conventional methods for the preparation of polyacrmide is suitable for generating simple shapes by mold casting but cannot easily form complex structures, such as cavities and curved passages useful for biomedical applications. The technology of 3D printing using stereolithography (SLA) or digital light processing (DLP), where polymerization is induced by localized, layer-by-layer illumination of photo-sensitive initiators,^[7,8] has the potential to fill this gap and dramatically expand the applications of polyacrylamide..

Polyacrylamide hydrogel is formed by copolymerizing acrylamide and N,N’ methylenebis(acrylamide) (referred to as bis-acrylamide) through free-radical polymerization, which is conventionally initiated by the free radical generator ammonium persulfate and catalyzed by tetramethylethylenediamine (TEMED). This is a slow reaction that typically requires 20–30 minutes to complete in a deoxygenated environment. While acrylates in organic solvents have been used widely for SLA or DLP,^[8] 3D printing of polyacrylamide must be conducted in an aqueous environment. In addition, the polymerization reaction must be sufficiently fast, to allow at least partial formation of each printed layer in no more than a few seconds. Successful 3D printing of polyacrylamide hydrogels was reported by dispersing a water-insoluble photo initiator, 2,4,6-trimethylbenzoyl-diphenylphosphine oxide, as sub-micron particles.^[9] However, the lengthy preparation procedure and the difficulty in post-printing removal of these particles raise concerns for biomedical applications. The resin also contained 8.9 wt% of a special tri-acrylate crosslinker, ethoxylated trimethylolpropane triacrylate (SR-9035), to substitute for the typical <0.5 wt% of bis-acrylamide. 3D printing has also been applied to polyethylene glycol diacrylate (PEGDA) using an efficient water-soluble photo initiator, lithium phenyl-2,4,6-trimethylbenzoylphosphinate (LAP),^{[10,11,12,13]} which has a broad absorption peak centered at 375 nm.^[12] The resulting structures showed a strong promise for vascular applications,^[13], but lacked some of the favorable properties of polyacrylamide such as its wide range of tunable stiffnesses and transparency for optical imaging.

To assess the use of LAP for 3D printing polyacrylamide, we substituted ammonium persulfate in the conventional reaction mixture with 1 mg ml⁻¹ LAP and observed that the polymerization reaction reached a steady state within seconds under a 15 W, 365 nm bench UV lamp, without either deoxygenation or TEMED. To allow layer-by-layer printing, we added a non-toxic water-soluble dye, ^[14] tartrazine, whose function was to confine the illumination within a thin layer through its strong absorption at the wavelength of 405 nm for SLA and DLP.^[15] The final resin for DLP therefore contained acrylamide, bis-acrylamide, LAP, and tartrazine (see Experimental Section for details). Note that both LAP and tartrazine are readily removable from the printed hydrogel by washing. For SLA, the scanning rate of the laser beam, typically at around 100 mm second⁻¹, proved to be too slow relative to the diffusion of photoactivated protomers for the formation of layered structures. This problem was overcome by including 50 vol% glycerol in the resin to increase the viscosity and decrease the rate of diffusion. Nevertheless, due to the limited viscosity of the resin and the negligible adhesion of polyacrylamide to the PDMS optical window, it was unnecessary to impose deadhesion vat movement between the layers as required for conventional DLP and SLA. The printing head instead moved up immediately upon the completion of each layer, with no indication of hindrance to head lifting or polymerized pieces left behind on the vat. This greatly increased the printing speed similar to continuous liquid interface production.^[8]

The performance of 3D printing was demonstrated qualitatively with a cubic frame structure (Figure 1a). Soaking the prints for 48 hours in water removed the yellow color of tartrazine, while the area of each side increased from 23×23 mm² to 28×28 mm² due to swelling (Figure 1b). A more complex test structure, which may serve as a prototype bioreactor, consisted of a spiral tube integrated into the inner space of a hollow cylinder (Figure 1c). Quantitative assessment of the X-Y resolution was performed with regularly spaced, alternating plateaus and trenches (Figure 1e). At a spacing of 150 μm, the structure was reproduced faithfully using an Autodesk Ember DLP printer (Figure 1f). At a spacing of 100 μm, plateaus became 131 μm in width while the trenches showed a width of 68 μm (Figure 1d), which reflected a common property of SLA and DLP as details approach the resolution of the optical system. Together, these experiments demonstrated the ability of 3D printed polyacrylamide to generate complex structures at resolution of 100 to 150 μm, as determined in part by the optical properties of the printer.

Figure 1.

Assessments of the performance of 3D printing of polyacrylamide hydrogels. Qualitative test involves a frame with interconnected segments of 5 mm × 2 mm × 2 mm and junctions of 2 mm × 2 mm × 2 mm (a, b). The overall size of the structure is 23 mm × 23 mm × 23 mm immediately after printing (a). Upon detachment from the printing head and soaking in water for 48 hours, the footprint increases to 28 mm × 28 mm due to swelling, while the yellow color of tartrazine disappears (b). A more complicated structure, which represents a prototype of water-permeable bioreactor, consists of a spiral tube integrated into the inner space of a hollow cylinder (c). The tube has an inner diameter of 4.2 mm and wall thickness of 1.4 mm. The tube is highlighted by filling the interior with a solution of 1 wt% bromophenol blue. Assessment of the resolution is conducted by printing a sheet 700 μm in thickness with alternating flat plateaus and trenches (e, CAD design of the pattern, 100 μm plateaus and trenches to the left, 150 μm plateaus and trenches to the right). Observation with a 10x dry phase contrast objective lens shows that 150 μm plateaus and trenches are printed accurately (f), while 100 μm plateaus and trenches appear as 131 μm-wide plateaus (dark regions) and 68 μm-wide trenches respectively (bright regions). The slanted texture in the prints is caused by diamond-shaped pixels of the projector used by the printer. Bar, 100 μm.

We then used 3D printing to create a polyacrylamide “chip” with micropatterned surface stiffness for testing the responses of cell migration to substrate rigidity. The approach addressed the lack of efficiency and consistency with previous fluidic methods for generating stiffness transitions.^[16] The chip consisted of a set of raised strips 500 μm in width and 300 μm in height, printed with 14.5 wt% acrylamide and 0.5 wt% bis-acrylamide for a high rigidity (Figure 2a and Figure 2b, grey regions). The space between the raised strips, also 500 μm in width, was immediately filled in after printing with a solution of 9.7 wt% acrylamide and 0.3 wt% bis-acrylamide to form a soft surface (Figure 2a and Figure 2b, light blue regions). Although the soft gel also covered the top of raised strips, its effect on surface stiffness was minimized by limiting the thickness to a single optical section (<6 μm based on a 10x/0.30 N.A. PlanFluor dry phase contrast objective lens;^[17] Figure 2c, inset). The compressive moduli as measured with a custom flat punch device were 8.96 kPa above the stiff strips and 3.97 kPa in between. While a similar approach based on mold-casting was used previously to create stiffness steps,^[18] the present approach increased the versatility given its ease of design modification and speed of printing.

Figure 2.

Fabrication of a polyacrylamide substrate with micropatterned stiffness for testing the responses of cell migration. Raised strips of stiff polyacrylamide 300 μm in height are created by 3D printing onto a glass coverslip (a, b, grey regions). The space between raised strips is then filled in with soft polyacrylamide (a, b, blue regions). which contains 0.2 μm fluorescent beads for visualization. While the surfaces of the raised strips are also covered with soft gel, its minimal thickness limits the effect on stiffness. The resulting substrate is therefore stiffer above the raised stiff strips than above the regions in between. Epifluorescence image of fluorescent beads shows strong fluorescence over the soft region (c, left), while the stiff region shows only very weak fluorescence from a thin layer of soft gel over the top (c, right). A single layer of beads over the stiff region becomes visible upon the enhancement of image brightness and contrast (c, inset). Scale bar, 50 μm. NRK-52E cells plated near the stiffness border (vertical lines, d-g) are tracked for 2.5 hours before and after contacting the border. The beginning and end of each cell track are indicated by green and red circles respectively. Cells at a low density turn away from the soft region when approaching a border from the stiff side (d), and enter the stiff region when approaching the border from the soft side (e). At intermediate densities of 100–500 cells/mm², the migration becomes more random regardless of the position relative to the stiffness border (f, g). Cells far away from the border show random migration regardless of surface stiffness (h, i).

We seeded the substrate with NRK-52E epithelial cells after functionalizing the surface with gelatin.^[19] At a low density, individual cells showed durotaxis as reported previously:^[16] cells turned away from soft surfaces as they approached the border from stiff surface (83.3%, p<0.0005; Figure 2d), while cells on soft surfaces migrated readily across the border onto stiff surfaces (93.3%, p<0.0005; Figure 2e). Cells away from the border migrated randomly (Figure 2h and Figure 2i). Durotaxis decreased when cells reached a density of 100–500 cells mm⁻¹,^[16] likely due to cell-cell interactions interfering with stiffness sensing (Figure 2f, Figure 2g and Figure 3a).

Figure 3.

Phase contrast time-lapse recording of NRK-52E monolayers near stiffness border. At an intermediate density, cells are able to cross the border between soft (right) and stiff (left) regions in either direction (a). At a high density, cells undergo striking collective segregation toward stiff regions, which causes the monolayer to rip apart and form a sharp boundary near the stiffness border (b). Stiffness border is indicated by dark lines. Elapsed times are shown above each panel. Arrows on panels at 6 hours indicate the direction of cell movement. Scale bar, 50 μm.

Unexpectedly, when cells reached a density of >650 cells mm⁻¹, the dense monolayer underwent striking collective segregation toward stiff regions, depleting cells from soft regions and forming sharp demarcation along the stiffness border (85.6%; p<0.0005; Figure 3b, Video S1). Distinct from durotaxis, which was observed only above the stiffness border and became obscured at intermediate cell densities (Figure 2f and Figure 2g), this collective segregation extended for hundreds of microns away from the border and showed a significantly higher speed of cell translocation (2.88 ± 0.17 μm minute⁻¹ versus 1.04 ± 0.05 μm minute⁻¹ for durotaxis or 0.72 ± 0.04 μm minute⁻¹ for random migration; Video S1). Therefore, it is unlikely to represent collective durotaxis. We suspect that the process was driven by the stronger traction forces on stiff surfaces than on soft surfaces,^[16] which then caused concerted contractile movement toward stiff surfaces.

In summary, the present method for 3D printing polyacrylamide is expected to greatly expand the use of this polymer for both biomedical and non-biomedical purposes. Beyond the present demonstration of lab-on-a-chip application for basic mechanobiology, 3D generated constructs may be used for diagnostic tests, artificial organs, and shaped fillers for tissue repair. Our results further suggest that stiffness-mediated segregation of cell collectives may play an important role in morphogenesis and that defined domains of stiffness may be engineered for guiding wound healing or artificial tissue formation.

Experimental Section

Preparation of acrylamide resin solution:

The resin for 3D printing consisted of acrylamide and bis-acrylamide (diluted 1:1 from a stock of 29 wt%:1 wt% acrylamide:bis-acrylamide mixture, Bio-Rad Laboratories, Hercules, CA), tartrazine (~0.027 wt%, absorbance at 425nm = 12.5, Sigma-Aldrich, St. Louis, MO), and LAP (0.1 wt%, Sigma-Aldrich). The solution was stored in a light-tight bottle at 4°C and brought to room temperature before printing.

3D printing and post-processing:

DLP 3D printing was carried out at room temperature with an Autodesk Ember 3D printer (San Rafael, CA). The parameters for printing was derived from those for Autodesk CMYK resins. Key parameters for a layer thickness of 100 μm were: no vat movement by setting all the rotation angles to 0; no z-axis overlifts of the head; 1 mm second⁻¹ for all the z-axis approach velocities; and 1.5 seconds for all the wait times before exposure. The exposure time was 8 seconds for the first layer and 7 seconds for burn-in layers and model layers. The same setting was used for a layer thickness of 25 microns except that the exposure time was 6 seconds for the first layer and 4 seconds for subsequent layers. SLA 3D printing was carried out at room temperature with a Peopoly Moai 130 printer (Hong Kong), whose laser was replaced with a 210 mW, 405 nm laser (Peopoly) running at a setting of 65. Tilting motion of the vat was inhibited by unplugging the motor from power source. The parameters for printing were: 100% infill density in a triangular pattern with 15% overlap; printing speed at 120 mm second⁻¹ except for an initial layer travel speed of 100 mm second⁻¹. Similar conditions should be applicable to other SLA/DLP printers of 405 nm with controllable exposure times and head/vat movements.

In most cases, printed objects were allowed to adhere directly to the printing head without supporting structures. When printing onto glass slides or coverslips, the glass surface was pre-treated with bind-silane according to the manufacture’s instructions (GE Healthcare, Waukesha, WI) before mounting onto a custom printing head degined to hold slides or coverslips. Printed polyacrylamide components were rinsed briefly with distilled water from a squirt bottle, then exposed to UV at 365 nm for 10 minutes under a 15W bench lamp (Cole Palmer, Vernon Hills, IL).

Cell culture substrates with micropatterned surface stiffness:

Substrates with patterned stiffness were prepared in a two-step process that involved printing stiff polyacrylamide gel with a raised pattern followed by paving the void space with soft polyacrylamide (Figure 2a and Figure 2b). The raised pattern consisted of parallel strips 500 μm from each other that were 500 μm in width and 300 μm in height, printed onto 45×50 mm² No. 1 glass coverslips (Fisher Scientific) treated with bind-silane.

The acrylamide solution for soft paving was prepared with a 1:2 dilution of acrylamide/bis-acrylamide (30 wt%, 29:1) with distilled water, LAP (0.15 wt%), and 1:1000 dilution of 0.2 μm fluorescent polystyrene beads (Molecular Probes, Carlsbad, CA). A volume of 30 μL was placed onto a clear plexiglass square and the glass coverslip with 3D printed gel was inverted on top immediately after printing. The assembly was exposed to 365 nm ultraviolet light at a distance of 2–3 inches from a 15 W bench UV lamp (Cole Parmer) for 20 minutes to polymerize the soft gel. The plexiglass square was then removed, and the gel was washed in distilled water overnight. The surface of the gel was conjugated with gelatin as described previously^[5,19]. Before use, the substrate was sterilized under 260 nm ultraviolet light for 30 minutes and, rinsed briefly with cell culture media.

Cell culture and microscopy:

NRK-52E rat kidney epithelial cells were obtained from American Type Culture Association (Manassas, VA) and maintained in Dulbecco’s modified Eagle’s medium (Life Technologies, Carlsbad, CA) supplemented with fetal bovine serum (10 vol% Thermo Scientific, Waltham, MA), streptomycin (50 μg mL⁻¹), penicillin (50 Units mL⁻¹), and L-glutamine (2 mм, Life Technologies, Carlsbad, CA). All cells were maintained in an incubator with 5% CO₂ at 37°C. NRK-52E cells were seeded onto polyacrylamide substrates and incubated for 4–5 hours before experiments to allow cell attachment.

Phase contrast images were taken every 30 minutes over at least 24 hours with a Nikon Eclipse Ti microscope equipped with a 10x/0.30 N.A. PlanFluor dry phase contrast objective lens and an on-stage culture chamber. Tracking at multiple sites was facilitated with a motorized stage and the Nikon Perfect Focus mechanism using custom software. Stiffness borders were identified through epifluorescence microscopy of the fluorescent beads added to the soft polyacrylamide gel. Responses of cell migration to substrate stiffness were characterized by visually tracking cell nuclei for 2.5 hours before and after the cell contacted a stiffness border from either the soft or stiff side. Observations of cells at a low density were performed with cells that did not divide and did not come within 20 μm of adjacent cells during the course of observation. For the analysis of cells at higher densities, the number of cells per unit area that approached the border from both soft and stiff regions was recorded. Statistical significance was determined through two-sample T-tests with p-values of 0.05 as the threshold of significance.

Supplementary Material

supp video1

Supporting Information

Video S1. Collective segregation of NRK-52E cells at a high density in response to a stiffness border. Time lapse recording reveals segregation of cells towards stiff substrates (left of the dark line), which eventually rips the monolayer apart to form a sharp boundary near the stiffness border. Stiffness border is indicated by a dark line. The first second of the video displays stiff and soft regions as revealed by fluorescent beads. Images are taken at 30-minute intervals. The total duration is 27 hours. Scale bar, 50 μm.